Case Study 24.2 — Ability-to-Repay: What the Rule Ended, and What It Cost

A real regulatory event, and a contested design choice. The Ability-to-Repay/Qualified Mortgage rule took effect on January 10, 2014. It is the most consequential underwriting mandate in the history of American residential lending, it ended an entire category of lending, and its original central threshold was so contested that its own authors built an exception large enough to swallow it and then, seven years later, removed it.

Borrower profiles below are labeled composites, assembled from documented pre-crisis product structures and from this book's constructed anchor files. No real borrower is described.

Background: what "underwriting" meant in 2006

Between roughly 2003 and 2007 a set of products spread through the market that shared one structural property: they did not require anyone to determine whether the borrower could make the payment out of documented income.

The mechanisms were distinct and they stacked.

• Stated income and reduced documentation. The borrower wrote an income figure on the application. Nobody verified it. These were originally designed for a narrow population — self-employed borrowers with complex returns — and migrated to wage earners who had pay stubs sitting in a drawer.
• Qualifying at the introductory rate. Hybrid adjustable-rate products carried a discounted rate for two or three years, then adjusted. Underwriting used the discounted rate. A borrower who could afford the first payment was approved for a loan whose later payments were substantially larger and had never been tested.
• Negative amortization. Payment-option products permitted a minimum payment smaller than accruing interest. The unpaid interest was added to principal. The balance grew while the borrower paid on time, and eventually the loan recast to a fully amortizing payment.
• Collateral-based approval. Underneath all of it sat an assumption: if the borrower could not pay, they would sell or refinance, because the house would be worth more. That assumption is not underwriting. It is a bet on a single variable, made simultaneously by everyone.

None of this required anybody to lie, which is the part worth sitting with. It required only that the question can this household make this payment out of documented income stop being the question the file was built to answer. The Financial Crisis Inquiry Commission's 2011 report and a substantial public record document the products and their performance.

Composite A — the 2006 approval. [labeled composite, assembled from documented pre-crisis product structures] A household stating \$9,000 a month of income, unverified. A 2/28 hybrid ARM at a 6.00% start rate with a fully indexed rate near 9.50%. Qualified at 6.00%. No reserves required. No documentation of assets. Approved in three days. The file contained no evidence of income, no evidence of assets, and no calculation of the payment the borrower would actually make in year three.

The issue: Congress writes a standard, and then draws a line

The Dodd-Frank Act added section 129C to the Truth in Lending Act, and the CFPB implemented it in Regulation Z § 1026.43. The core is a standard, not a formula: a creditor must make a reasonable and good faith determination, at or before consummation, that the consumer has a reasonable ability to repay the loan according to its terms — considering eight enumerated factors and verifying income, assets, and obligations from reasonably reliable third-party records.

Read against Composite A, every mechanism above is dead. Stated income fails the verification requirement. Qualifying at a teaser rate fails the payment-calculation rule, which requires the greater of the fully indexed rate or the introductory rate. Negative amortization is a product feature no Qualified Mortgage may have. And the collateral assumption is directly excluded: the rule requires consideration of income or assets other than the value of the dwelling that secures the loan.

That is the achievement, and it is enormous. It is also, on its own, a litigation standard — and lenders do not originate into an undefined standard at scale. So Congress and the Bureau defined a category of loans presumed to satisfy it: the Qualified Mortgage.

And here the design became contested. The original General QM definition contained a bright line: a debt-to-income ratio not exceeding 43 percent, with income and debt calculated under a rigid appendix to the regulation (Appendix Q).

Two problems appeared immediately.

Problem one: 43 is a number, and ability to repay is not. A household earning \$4,000 a month at a 42% ratio has roughly \$2,320 left for everything that is not debt. A household earning \$18,000 a month at a 46% ratio has roughly \$9,720. The second household is in a materially stronger position and the bright line declines it. Residual income — the measure the VA has used for decades, and the measure that appears in the QM rule's own rebuttable-presumption standard — captures this. A ratio does not. Chapter 8 made this argument on its own terms; the ATR rule made it a legal problem.

Problem two: the rule immediately exempted most of the market from its own line. The Temporary QM category — universally called the GSE Patch — provided that loans eligible for purchase or guarantee by Fannie Mae or Freddie Mac while in conservatorship were Qualified Mortgages without regard to the 43% ceiling, subject to the other requirements. The Patch was scheduled to expire on a date certain or when the enterprises left conservatorship, whichever came first.

Consider what that structure conceded. The Bureau wrote a bright-line DTI test for General QM, and in the same rule provided that an enormous share of conventional originations need not meet it — because an automated underwriting system evaluating the whole file was a better predictor than a single ratio. The exception was an admission about the rule.

Problem three: Appendix Q could not describe real income. The documentation standards for General QM income were rigid and, for self-employed borrowers in particular, frequently produced a number that no competent underwriter would have used and that no investor guideline required.

Composite B — the borrower the line declined. The Fulton Avenue file. [constructed teaching file] A self-employed contractor, S-corporation, six employees, two years of returns. The Fannie Mae cash-flow analysis produces \$9,020.83** as a 24-month average and **\$8,916.67 using the most recent year alone; because income declined 2.3% year over year, the underwriter uses the lower figure. The borrower's accountant told them they "make about \$9,500 a month."

Three defensible numbers — \$9,500, \$9,020.83, \$8,916.67 — and the ratio the file is judged by moves depending on which one applies. A 43% bright line asks a documentation methodology to carry a precision it does not have. That is the criticism, and it is a fair one.

What it shows

The rule ended a category of lending and the ending held. This is worth stating plainly, because regulatory interventions frequently do not work and this one did. Stated income on owner-occupied consumer mortgages did not survive. Negative amortization did not survive as a QM feature. Qualifying at a teaser rate did not survive. More than a decade later, none of them have returned to the covered market, and the reason is structural rather than cultural: ATR liability cannot be sold away, and a violation can be raised defensively in foreclosure years later by way of recoupment or setoff. No investor buys that.

The bright line did not hold, and the Bureau eventually said so. In December 2020 the Bureau removed the 43% DTI limit from the General QM definition and replaced it with a price-based test keyed to the loan's APR relative to the average prime offer rate for a comparable transaction, with the permitted spread varying by loan amount and lien position. Appendix Q went with it. The temporary GSE-eligibility category was allowed to expire. A separate Seasoned QM category was created, under which a portfolio loan meeting product restrictions and performing over a defined seasoning period attains safe-harbor status.

The reasoning behind the substitution is worth understanding, because it is not obvious. Price is information. A loan priced far above the market for comparable transactions is a loan the market has independently judged to be riskier — through underwriting, credit, documentation, and every other input a rate reflects. Using price as the proxy is an admission that the market's composite judgment about a file is more informative than one ratio computed under one methodology.

But the analysis did not disappear. This is the part practitioners over-read. The amended General QM definition retains the requirement that a creditor consider the consumer's income or assets, debt obligations, alimony, child support, and monthly debt-to-income ratio or residual income, and verify them from reasonably reliable third-party records. The ceiling is gone. The obligation to compute, consider, and document is not.

And the credit box narrowed regardless. The honest accounting of ATR/QM includes a cost. Borrowers with genuine ability to repay whose income does not fit standard documentation — self-employed borrowers, borrowers with substantial assets and modest income, borrowers with irregular but reliable earnings — became harder to serve. The market's answer was to rebuild a documented alternative: non-QM lending, in which ATR applies in full, income is verified from third-party records such as bank statements or asset schedules, and the creditor accepts the absence of a QM presumption in exchange for underwriting flexibility. That is Chapter 34's subject, and it is the correct answer — but it took years to rebuild, it prices higher, and in the interval real borrowers went unserved.

Outcome

Where the rule sits today, stated structurally:

Do not print the current price thresholds from memory, from a study guide, or from this book. Look them up in Regulation Z § 1026.43(e) and the CFPB's current published figures.

The lesson

Three, in order of how often they matter to a loan officer.

1. The rule is why your file looks the way it does. Every verification in Chapters 10 through 12 — the pay stubs, the W-2s, the written verification of employment, the two months of statements, the sourcing of a \$4,900 deposit — is not investor bureaucracy alone. It is federal law requiring the creditor to verify from a third party. When a borrower asks why their own bank account is not sufficient evidence of their own money, that is the answer, and it is a better answer than "it's just what underwriting wants."

2. A bright line is a tool, not a truth — and this is the book's own recurring argument arriving in statute. The 43% ceiling declined borrowers who could afford the house and approved borrowers who could not, which is exactly what this book says about debt-to-income generally. The rule's own authors built an exception around it and then removed it. Learn the structure; treat any specific threshold as perishable.

3. Neither the old rule nor the new one is the constraint that will actually decide your file. The Linden Street file's back-end ratio is 42.66%. Under the old General QM it cleared the ceiling by 34 basis points — uncomfortably close for a file that would soon absorb a \$611.00 monthly obligation on day 44 and jump to 48.48%. Under the current definition there is no ceiling to clear. And the loan is still dead at 48.48% as originated, for two reasons that have nothing to do with QM: Fannie Mae's eligibility and the lender's overlays impose their own limits, and a creditor sitting on a credit refresh that shows a new debt can no longer claim it made a reasonable, good-faith determination from current verified obligations. Chapter 19 owns the resolution — paid in full, zero-balance letter, findings re-run, four business days.

QM is a legal category. Salability is a business decision. Ability to repay is a question about a household. A loan officer needs all three, and confusing them is how a file dies at day 44.

Discussion questions

1. The original General QM used a 43% DTI ceiling; the current one uses a price-based test. Argue for each design. Which better serves a borrower with a 46% ratio and eighteen months of reserves? Which better serves a regulator trying to supervise a thousand lenders at once?

2. The GSE Patch let a large share of conventional originations skip the 43% ceiling for years. Was that a pragmatic bridge or an admission that the ceiling was the wrong instrument? What would you need to know to answer honestly?

3. Composite B has three defensible income figures — \$9,500, \$9,020.83, and \$8,916.67 — depending on methodology. Explain how a bright-line ratio test interacts with that ambiguity, and whether the price-based test resolves it, relocates it, or ignores it.

4. Non-QM lending exists because ATR applies to loans that cannot meet a QM definition. Explain what the lender gives up by originating non-QM, what the borrower gets, and why "non-QM" and "no-documentation" are not the same thing — in the words you would use with a borrower who has heard the term and is nervous.

5. An ATR violation may be raised defensively in a foreclosure by recoupment or setoff without regard to the usual limitations period. Explain what that does to the economics of buying a mortgage loan, and connect it to why a HOEPA high-cost loan is effectively unsalable.

6. The rule ended stated-income lending on covered consumer mortgages and it also made some creditworthy borrowers harder to serve. Both are true. Write the two-paragraph honest assessment you would give a policymaker — one paragraph on what was gained, one on what was lost — without inventing a statistic.